TITLE:
Hidden Markov Models to Estimate the Lagged Effects of Weather on Stroke and Ischemic Heart Disease
AUTHORS:
Hiroshi Morimoto
KEYWORDS:
Hidden Markov Model, Self Organized Map, Stroke, Cerebral Infarction, Ischemic Heart Disease
JOURNAL NAME:
Applied Mathematics,
Vol.7 No.13,
August
4,
2016
ABSTRACT: The links between low temperature and the
incidence of disease have been studied by many researchers. What remains still
unclear is the exact nature of the relation, especially the mechanism by which
the change of weather effects on the onset of diseases. The existence of lag
period between exposure to temperature and its effect on mortality may reflect
the nature of the onset of diseases. Therefore, to assess lagged effects
becomes potentially important. The most of studies on lags used the method by
Lag-distributed Poisson Regression, and neglected extreme case as random noise
to get correlations. In order to assess the lagged effect, we proposed a new
approach, i.e., Hidden Markov Model by Self Organized Map (HMM by SOM) apart
from well-known regression models. HMM by SOM includes the randomness in its
nature and encompasses the extreme cases which were neglected by
auto-regression models. The daily data of the number of patients transported by
ambulance in Nagoya, Japan, were used. SOM was carried out to classify the
meteorological elements into six classes. These classes were used as “states”
of HMM. HMM was used to describe a background process which might produce the
time series of the incidence of diseases. The background process was considered
to change randomly weather states, classified by SOM. We estimated the lagged
effects of weather change on the onset of both cerebral infarction and ischemic
heart disease. This fact is potentially important in that if one could trace a
path in the chain of events leading from temperature change to death, one might
be able to prevent it and avert the fatal outcome.